Optimization of neighbor list techniques in liquid matter simulations

The performance of neighbor list techniques in molecular dynamics simulations depends on a variety of parameters, which may be adjusted for maximum efficiency. Here, a model is presented which allows to choose optimal parameters for the performance of Verlet- and linked-cell lists. In several cases, an efficiency gain of ≈ 50% is found if parameters are chosen adequately. Test cases are presented for Lennard-Jones systems at different state points. Good agreement between analytical model and simulation results is found.